ReadyHermes
What it doesAppsPricingBot toolsFAQ
PricingLog inStart free trial
On this page
  1. What self-hosting actually involves
  2. Why an always-on agent needs a machine that stays up
  3. The ongoing chores
  4. What managed hosting takes off your plate
  5. What you still control when it is managed
  6. Side by side
  7. Choose self-hosting if…
  8. Choose managed if…
  9. Taking your agent’s files with you
  10. How to decide
  11. Questions
Start free trial

Hermes Agent hosting

Hermes Agent Self-Hosting vs Managed Hosting: Which One Fits You?

Updated 10 October 2026

Hermes Agent is open source, so you can run it yourself. You can also pay someone to run it for you. Both are reasonable. They suit different people.

This page lays out what each path involves, where each one is the better choice, and where the work hides. We run a managed service, so we have a stake in the answer. We have tried to be fair anyway, and in several cases self-hosting is simply the right call.

ReadyHermes is an independent hosting service. It is not affiliated with Nous Research, the team that makes Hermes Agent.

What self-hosting actually involves

The install itself is short. The upstream repo gives a one-line installer for Linux, macOS and WSL2:

curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash

On native Windows, use PowerShell instead:

iex (irm https://hermes-agent.nousresearch.com/install.ps1)

Then you tell it which AI model to use and run the setup wizard:

hermes model
hermes setup

hermes model is where you pick a provider and a model. Nous Portal, OpenRouter, OpenAI, your own endpoint and many others are options. hermes setup walks through the rest.

To reach the agent from a chat app, you set up the messaging gateway and start it:

hermes gateway setup
hermes gateway

For Telegram you create a bot with BotFather, then either run hermes gateway setup or set TELEGRAM_BOT_TOKEN and TELEGRAM_ALLOWED_USERS (your numeric user ID) yourself. Discord, WhatsApp and other platforms go through the same gateway.

How long that takes depends on how at home you are in a terminal. Self-hosting is free software. You pay separately for the machine and for model usage.

Why an always-on agent needs a machine that stays up

You can run Hermes on a laptop and it will work fine while the lid is open. The trouble starts when you want the things that make an agent useful: a morning briefing at 8am, a reminder on Friday, a page it checks while you sleep, a Telegram bot that answers when your laptop is in a bag.

Scheduled jobs and chat apps only work while the gateway process is running. In practice that means a server or VPS that is on all the time and that you look after. A home machine can do it too, if you are happy to keep it powered, connected and restarted after power cuts.

This is the part people tend to underestimate. The install is quick. Keeping it running for months is the actual job.

The ongoing chores

Once the agent is live, these become yours:

  • OS updates and restarts. Security patches, reboots, and checking the gateway came back afterwards.
  • Hermes updates. Pulling new versions and reading what changed.
  • Backups. Your agent’s memory, skills and files live on that disk. If the disk dies and you never copied it, they are gone.
  • Security. Firewall, SSH keys, who can message the bot, and keeping API keys out of the wrong places.
  • Model keys and bills. You hold the accounts with your AI providers, watch the usage and pay the invoices.
  • Noticing when it breaks. If the gateway stops at 3am, nobody tells you. You find out when a reminder does not arrive.

None of this is hard for someone who runs servers already. For someone who does not, it is a steady drip of small tasks, and each one is a chance to get stuck.

What managed hosting takes off your plate

With managed Hermes Agent hosting such as ReadyHermes, you sign up and get your own Hermes Agent. There is no server to rent, no Docker, and no API key to find. The agent is ready about a minute after signup. Each account gets its own dedicated agent with its own memory and skills, isolated from other users.

What you stop doing:

  • Renting, securing and patching a machine.
  • Installing and updating Hermes.
  • Choosing and paying for a model. AI usage is included on both plans (the model is DeepSeek V4.1 Flash, with fair-use limits).
  • Setting up chat apps from scratch. Telegram, WhatsApp, Discord, WeChat and Feishu / Lark connect from the dashboard. Most take about a minute; the first WeChat connection takes 2 to 4 minutes.
  • Watching for outages. The service is monitored for you 24/7, and updates are handled for you.

You still get the agent’s built-in abilities: long-term memory, skills it writes for itself, scheduled jobs, web search, a real web browser and its own Linux computer. Connected apps like Gmail, Google Calendar, Notion, Slack and GitHub are set up by asking the agent in chat.

Pricing is monthly. Lite is $5 a month (1 vCPU, 2 GB memory, 10 GB storage). Pro is $10 a month (2 vCPU, 4 GB memory, 40 GB storage). There is a 30-minute free trial with no card. See pricing for details, or cheap Hermes Agent hosting for how it compares on cost.

What you still control when it is managed

A common worry is that managed means a locked box. It is less locked than it sounds. The dashboard has a section called Bot tools with these tabs:

  • SSH. Paste a one-time auth key from your own Tailscale account and your agent joins your private network. Bot tools shows the exact command, for example ssh root@<its name>, and you get root. No public port is opened.
  • API. Turn on an OpenAI-compatible endpoint with a key, and call your agent from your own scripts or servers. It is meant for server-to-server use, not browser web apps.
  • Backup and Restore. Download your agent as a .zip, and load a backup back later. Managed settings such as the AI model and connected chat apps are not restored.
  • Files. Browse, upload, download, rename and delete. There is no in-browser editor.
  • Power, Desktop and My computer. Start, stop or restart the agent, watch its desktop and browser live, or let it work with chosen folders on your own computer. My computer is read-only by default, and saving files and running commands are separate permissions that stay off until you turn them on.

On paid plans you can also bring your own model key from providers such as OpenRouter, OpenAI, Anthropic, Google Gemini, DeepSeek, Groq and others, or any OpenAI-compatible endpoint with a public https address. Your provider then bills you directly, and you can switch back to the included AI whenever you like.

What you do not get is control of the host machine itself. You cannot choose the hardware, and a model you run at home only works if you expose it at a public https address.

Side by side

FeatureSelf-hostingManaged (ReadyHermes)
Setup timeDepends on your terminal skills, plus time for chat apps and a server.About a minute after signup. Chat apps connect from the dashboard.
Server upkeepYours: OS updates, Hermes updates, restarts, security.Handled for you. Monitored 24/7.
Model keys and billsYou choose a provider, hold the keys and pay the usage.AI included (DeepSeek V4.1 Flash, fair-use limits). Optional own key on paid plans.
Chat appsTelegram, Discord, WhatsApp and others through the gateway, set up by you.Telegram, WhatsApp, Discord, WeChat and Feishu / Lark from the dashboard.
UptimeAs good as your machine and how quickly you notice a problem.Monitored for you 24/7.
Control and customisingTotal. Any model, any hardware, any change to the code.Root SSH over Tailscale, an API, backups and files. Not the host machine.
Cost shapeFree software. You pay for the machine and for model usage, separately and variably.$5 (Lite) or $10 (Pro) a month, billed monthly, cancel any time.
Data locationWherever your machine is.Account and agent data stored in the EU. Messages go to the model provider to write a reply, which may be outside the EU.

The managed column describes ReadyHermes only. Other managed hosts differ. For one comparison, see ReadyHermes vs Hostinger.

Choose self-hosting if…

  • You want full control. You want to read the code, change it, and run exactly the version you choose.
  • You want to run models on your own hardware. A local model on your own GPU that stays on your private network is a self-hosting job. A managed service can only reach model endpoints with a public https address.
  • You like tinkering. If setting up the box is part of the fun, paying someone to skip it takes the fun away.
  • You already run servers. If you have a VPS or a home server with spare room and a habit of patching it, one more service adds little work.
  • You need the data on a specific machine. Some people need files to stay in one place they control.
  • You want to pay only for what you use. A cheap machine plus a pay-per-token key can cost less than a subscription if you use the agent lightly.

Choose managed if…

  • You want an agent today, not a project. About a minute from signup to a working chat box.
  • You do not want to run a server. No patching, no firewall, no checking whether the gateway survived a reboot.
  • You want one predictable bill. A flat monthly price with the AI included, instead of a machine bill plus a token bill that moves.
  • You want chat apps without the plumbing. Including WeChat and Feishu / Lark, which take extra setup on your own.
  • Your agent is for work, and others rely on it. Reminders and briefings should arrive even when you are busy with something else.
  • You want to try before committing. The free trial gives you 30 minutes of Lite with no card. If you do nothing after it ends, the agent pauses, and you have 24 hours to choose a plan before it is deleted.

You do not have to pick forever. You can start managed to see what an always-on agent is good for, then move to your own server once you know what you need.

Taking your agent’s files with you

Leaving a managed service is a fair worry. On ReadyHermes, the Backup tab downloads your agent as a .zip while your trial or plan is active. Tick Conversations, Workspace files and memory, and Skills to include them; the default backup holds only settings and instructions. That gives you your agent’s files, so you are not locked in. We are not claiming the .zip drops straight into a self-hosted install. Treat it as your own copy of the files, to keep or to read through.

You can also download a copy of your data or delete your account yourself from Privacy & data in the dashboard.

How to decide

If you are unsure, ask which cost you mind more: your time or a few dollars a month. If time, go managed. If the dollars, or the control, go self-hosted.

To see what the managed route feels like, the step-by-step guide walks through signup to a first working task, and you can start the free trial in a minute. If you would rather self-host, the Hermes Agent repo has everything you need.

Try your own Hermes Agent free for 30 minutes.

No card, no server, no API key. Your agent is ready in about a minute.

Start free trialSee pricing

Questions

Is Hermes Agent free to self-host?

The software is free and open source. You still pay for the machine it runs on and for whatever AI model usage it consumes, and those costs are separate and depend on what you pick.

Can I run Hermes Agent on my laptop?

Yes, and it works while the laptop is on and awake. Scheduled jobs and chat apps only respond while the gateway is running, so for an agent that answers all day you need a machine that stays up, usually a VPS.

What do I need to self-host Hermes Agent?

A Linux, macOS or WSL2 machine (or Windows with PowerShell), the one-line installer, and an account with an AI model provider. After install you run hermes model, hermes setup, and for chat apps hermes gateway setup followed by hermes gateway.

Do I get root access on a managed ReadyHermes agent?

Yes, through the SSH tab in Bot tools. You paste a one-time auth key from your own Tailscale account, your agent joins your private network, and you connect with the command shown there. No public port is opened.

Can I use my own model key on managed hosting?

On paid plans, yes. In the Model key section you can add a key from providers such as OpenRouter, OpenAI, Anthropic, Google Gemini and DeepSeek, or any OpenAI-compatible endpoint with a public https address. Your provider bills you directly, and you can switch back to the included AI at any time.

Can I leave ReadyHermes and take my agent with me?

While your trial or plan is active you can download your agent as a .zip from the Backup tab in Bot tools (tick Conversations, Workspace files and memory, and Skills to include them), and download a copy of your data from Privacy & data. We do not claim the .zip imports directly into a self-hosted install, but the files are yours.

Is ReadyHermes made by Nous Research?

No. Hermes Agent is made by Nous Research. ReadyHermes is an independent hosting service and is not affiliated with them.

Keep reading

  • How to use ReadyHermes, step by step
  • Managed Hermes Agent hosting
  • Cheap Hermes Agent hosting
  • ReadyHermes vs Hostinger for Hermes Agent hosting
ReadyHermes

Hosted Hermes Agent for everyone.

ProductWhat it doesPricingFAQHow to useHermes hosting
AccountSign upLog inSupport
LegalPrivacyTermsContact
Featured on 50+ directories
Powered by ProductDirsFeatured on ShowMeBestAIFeatured on AIJustBetter.comFeatured on Twelve ToolsFeatured on Wired BusinessReadyHermes - Featured on Startup FameOpenHunts Club MemberLaunchpadly Startup DirectoryFeatured on EarlyHuntFeatured on ToolhunterReadyHermes - Featured on AI Agents DirectoryFeatured on ToolPilot.aiAgentHunter BadgeAgentHunterFeatured AI AgentFeatured on Launch Llama ToolsFeatured on the Postmake directoryListed on IndieToolsFeatured on FoundrListReadyHermes on Launch & LoopReadyHermes — Featured on Smol LaunchReadyHermesReadyHermes on LaunchNestListed on Kick ProductListed on Uno DirectoryReadyHermes on Nick LaunchesFeatured on Better LaunchVerified on DANG!Featured on aihuntlist.comFindYourAgent LogoStork Verified — stork.ai AI tools directoryFeatured on DeepLaunch.ioFeatured on Dofollow.ToolsFeatured on LaunchVaultFeatured on LaunchKiwiFeatured on neeed.directoryAI X CollectionAcid ToolsAI Tech ViralWe Like ToolsVorgestellt auf Beste KI ToolsVorgestellt auf KI Tools HeuteFeatured on WhatsthebigdataHuzzler Embed BadgeSubmit AI ToolsReadyHermes | AidirsFeatured on NavifyAI.com - AI Tools DirectoryFeatured on Software on the WebFeatured on Next Launch

Also listed on Featured on SeeWhatNewAI · Featured on NewAIForYou · AI Toolz Dir · AI Tool Trek · AI Hustle Tools · ReadyHermes on Launchtory

Hermes Agent is open-source software by Nous Research. ReadyHermes is an independent hosting service and is not affiliated with Nous Research.

© 2026 ReadyHermes. All rights reserved.